The incidence of non-Hodgkin's lymphoma (NHL) has nearly doubled over the past four decades in the US and most westernized countries, and changes in disease reporting and classification and in the prevalence of known risk factors do not account for the entirety of this increase. Two prior studies provided evidence for an association between use of tricyclic antidepressants (TCAs) and the occurrence of NHLs, but they were unable to investigate associations within NHL subtypes, which are known to be etiologically diverse. In response to the National Institutes of Health announcement for small grants in cancer epidemiology, we propose to conduct a case-control study of NHL among members of Group Health (GH), a large integrated healthcare delivery system in western Washington State. Cases will include all GH members with a diagnosis of NHL between 1980-2009, who were e25 years old and had been GH members for e2 years at diagnosis, and who meet the exclusion criteria. Controls will be selected from the defined population of GH members, thus they should be representative of the population that gave rise to cases. They will be matched 8:1 with cases on age (5-year age groups), sex, date of case's diagnosis (the reference date), and length of membership. TCA use will be ascertained using a 30-year automated pharmacy database which contains detailed data on every prescription dispensed at GH, recorded consistently and prospectively throughout the study period and thus prior to reference date and similarly for cases and controls. We will test whether: (1) prior TCA users are at an increased risk of NHL, and (2) whether the association varies by type of NHL. These data will permit a valid assessment of the risk of NHL in relation to the duration, dose, regency, and type of TCA use, both overall and for specific types of NHL. The proposed study has the advantages typical of the case-control design, as it is efficient in terms of both costs and timeline, while also avoiding many of the biases often found in case-control studies. The use of prospectively collected automated data prevents recall bias and information bias, and use of the defined population of GH enrollees ensures that controls are representative of the population at risk of NHL. PUBLIC HEALTH RELEVANCE: Rates of non-Hodgkin's lymphoma (NHL) have nearly doubled over the past four decades in the US, making it the sixth most common cancer in both men and women, and this increase is not entirely explained by known factors. Previous studies suggest that tricyclic antidepressant (TCA) users are at an increased risk of NHL, but they did not examine the risk of specific NHL subtypes in association with TCA use. Because TCAs continue to be widely used, and because of the etiologic heterogeneity across subtypes in this common class of cancers, an assessment of NHL incidence in TCA users represents an important undertaking.